邻域多粒度粗糙集知识更新增量算法  

Incremental Algorithm for Knowledge Updating of Neighborhood Multigranulation Rough Set

作　　者：徐怡[1,2] 孙伟康 王泉 XU Yi;SUN Wei-kang;WANG Quan(Key Laboratory of Intelligent Computing and Signal Processing,Ministry of Education,Anhui University,Hefei 230039,China;College of Computer Science and Technology,Anhui University,Hefei 230601,China)

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室,合肥230039 [2]安徽大学计算机科学与技术学院,合肥230601

出　　处：《小型微型计算机系统》2020年第5期908-918,共11页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61402005)资助;安徽省自然科学基金项目(1308085QF114)资助;安徽省高等学校省级自然科学基金项目(KJ2013A015)资助;国家留学基金委员会项目(201606505034)资助.

摘　　要：邻域多粒度粗糙集由邻域关系族组成,能够有效地处理连续的数值型数据,所以有广泛的应用前景.但在实际应用中,论域的变化会导致多粒度环境下粒度划分的变化,进而导致原始知识的变化,如本文中邻域多粒度粗糙集的正域、负域和边界域的变化.为了解决这一问题,本文提出了一种基于矩阵的增量式知识更新方法.首先,我们给出了邻域多粒度粗糙集模型的矩阵表示方法.在此基础上,提出了正域、负域和边界域的增量更新算法.通过使用该方法,减少了知识更新的时间复杂度,提高了算法的效率,通过具体的数据集和实验验证了该算法的有效性.Neighborhood multigranularity rough set is composed of neighborhood relation family,which can process continuous numerical data effectively,so it has a wide application prospect.But in practical application,the change of universe will lead to the change of granularity division in multigranularity environment,and then lead to the change of original knowledge,such as the change of positive region,negative region and boundary region of neighborhood multigranulation rough set.In order to solve this problem,this paper proposes an incremental knowledge updating method based on matrix.Firstly,we give the matrix representation of neighborhood multigranularity rough set model.On this basis,the incremental updating algorithms of positive region,negative region and boundary region are proposed.By using this method,the time complexity of knowledge updating is reduced,and the efficiency of the algorithm is improved.The effectiveness of the algorithm is verified by specific data sets and experiments.

关 键 词：邻域多粒度 知识更新 矩阵 增量方法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]